Net Promoter Score (NPS) measures overall customer loyalty toward a brand. By subtracting the number of Detractors from Promoters (two consumer categories) you’ll get the NPS score, which indicates how consumers feel about your performance and overall brand.
NPS is thought of as the gold standard metric that measures customer satisfaction. First, it’s used to collect quantitative data from one simple question, along the lines of:
“On a scale of 0 to 10, how likely are you to recommend us to a friend or colleague?”
And classifies respondents into three consumer categories based on their scores:
Calculating the NPS score is simple: you subtract the percentage of Detractors from the percentage of Promoters (Passives are always excluded). For example, if 20% of respondents are Detractors, 30% are Passives, and 50% are Promoters, the NPS score is 50 - 20 = 30.
The next part of NPS includes a follow-up question, asking customers to expand on their scores. This is the qualitative NPS score, if you like, which we’ll go into more detail about below.
Although the NPS score is valuable to your brand or product/service, you can gain even deeper insights from the qualitative NPS responses. By asking an open-ended question, customers are able to communicate the reasons for their initial score.
For example, knowing why Detractors give negative responses can help you address specific pain points in your company. Or, knowing why Promoters leave scores of 9 or 10 offers insight about what your most loyal customers’ value or like about your product.
However, open-ended NPS answers are more complex to analyze than NPS scores. You’ll need time and resources to go over each response and collect insights. You’ll also need to question the accuracy of those insights – customer service agents that are left to sift through large amounts of responses will grow tired, leaving them more prone to errors.
While there are no hard and fast rules about how to analyze your open-ended NPS responses, there are some best practices you can follow.
Below are several best practices you can follow to get the most out of your unstructured (or qualitative) NPS data:
As former Director of Product at LinkedIn Sachin Rekhi eloquently puts it, “the most actionable part of the NPS survey is the categorization of the open-ended comments from Promoters and Detractors.”
This means going through open-ended responses and categorizing each one into predefined topics, such as Functionality, Usability, or Reliability. You could also tag each open-ended response with a sentiment: Positive, Negative, or Neutral.
Categorization allows you to quantify the results of the open-ended responses and provide context to your NPS scores. For example, you might discover that 90% of your Detractors complain about your customer service. Or that most of your Promoters love your usability. It’s these results that become actionable insights, helping you make data-driven decisions that improve your brand and product or service.
Demographic data, such as age, gender, location, income, etc, helps further separate customers into groups so you can find out which demographic has more Promoters or Detractors.
You could also pinpoint aspects that appeal to different age groups. Do younger Detractors prefer more personalized customer experiences? Or, is there a particular feature that younger Promoters love that those in older age groups don’t value as much?
These insights can help you create more customized experiences, and make decisions based on your target group.
By analyzing answers to the follow-up question in NPS you can also discover root causes, in other words, what is causing an issue.
For example, if your laptop fails you take it to be serviced to ‘root out’ the cause of the failure. Usually, you’ll receive a survey asking you to rate the level of service and if you would recommend it to a friend or colleague. The pool of answers to this question might alert you to a recurring problem that needs to be addressed right away.
So, how does root cause work?
Well, there is no right way to do root cause analysis, but one of the most popular techniques is the 5 Whys approach. Not all problems stem from a single root cause, which is why this iterative technique helps get to the bottom of issues with 5 rounds of ‘Why’ questions. The answer to the first ‘Why’ question is the basis for the next questions.
Here’s an example of how you can use it to find the root cause of a mobile app that is full of bugs:
NPS survey open-ended response from A Detractor: “The app is full of bugs.”
1st ‘Why’: a new update was recently deployed.
2nd ‘Why’: customers were requesting new updates.
3rd ‘Why’: the previous version did not include a key feature.
4th ‘Why’: budget constraints prevented developers from including the key feature.
5th ‘Why’: the feature wasn’t properly documented
Result of the 5 ’Whys’ Root Cause Analysis: The original feature was not properly documented.
With these insights, you can address the issue by sharing them with relevant teams to ensure it doesn’t happen again. From enforcing standardized documentation policies to hiring more staff, you can take action and fix issues.
Tips to identify the root cause of customer feedback include:
It’s important to track results from survey responses and search for trends over a period of time.
SaaS companies, for example, are advised to send surveys every quarter – that’s one every 90 days – which will gauge how customer behavior and opinion changes from one quarter to the next.
Let’s say you notice an increase in the number of responses mentioning bugs, so you hire more developers to keep bugs to a minimum in the next quarter. You can check to see if your investment paid off by sending out NPS surveys in the following quarter, and comparing the results to the previous quarter.
Retail companies and e-commerce sites are advised to send a survey after each purchase to assess how the customer feels about their purchase, their experience, and the overall brand.
This systematic request for feedback will help you detect sentiment, and identify if customers respond well to new products, service features or functionality. Furthermore, you can listen to what customers need from your brand and uncover trends. Maybe they need more documentation or a more intuitive user experience, maybe a chatbot could help the customer experience, or perhaps pricing plans are through the roof!
In short, analyzing feedback over time helps you pinpoint what customers like or dislike about your business and stay on top of problems that may arise.
Once you have actionable insights from your NPS analysis, what should you do to implement change?
One simple action you can take is sharing NPS results with the relevant teams in your company. For example, if customers are complaining about bugs in your mobile app, NPS results should be shared with your product team to address the problem.
Or, if your customers want more comprehensive subscription plans that are affordable, then sharing NPS results with finance and product teams can help them adapt plans to suit your customers’ needs.
NPS analysis helps you understand your customers on a deeper level, and passing on insights to relevant teams is the most effective way to solve any problems and discover opportunities that can help improve customer loyalty.
By using machine learning, businesses are able to analyze NPS results like never before – automatically, consistently, at scale, and in real-time – delivering a superior level of understanding about your customers.
Imagine you work at Walmart and need to sift through thousands of NPS responses. You could go through this data manually but you’re likely to end up with unreliable results. Plus, it’s virtually impossible to get through all this data while it’s still fresh.
Instead, machine learning models can be trained to understand the sentiments, topics, or urgency behind your NPS text data, so you can gain valuable and reliable insights in minutes.
There are plenty of machine learning tools online that can help you automatically analyze NPS responses. MonkeyLearn, for example, offers various text classifiers that can categorize NPS responses into pre-defined groups according to their content. Some of the most popular text classifiers are:
Topic analysis interprets and categorizes large collections of text according to individual topics or themes. For example:
“The website’s chatbot was truly engaging – it walked me through the entire process of subscribing for extended support. Easy and simple to do!”
A topic analysis model can be trained to automatically tag this as Chatbot and Ease of Use. Try topic analysis out for yourself by using MonkeyLearn’s very own NPS SaaS feedback classifier. This pre-trained model will group feedback into Customer Support, Ease of Use, Features, and Pricing.
Sentiment analysis automatically detects the emotional undertones of customer reviews, NPS survey responses, social media posts, and so on, which helps businesses understand how their customers feel about their brand, product, or service.
For example, you can train a sentiment analysis model to process the following review and categorize it as Positive:
“I love my new website made by Wix! The drag-and-drop features were very intuitive.”
Check out this pre-trained sentiment analysis model to see how feedback is categorized into Positive, Neutral and Negative.
Used to automatically identify customer intentions within text. For example, a customer who leaves feedback stating they will downgrade if prices go up can be automatically tagged by an intent classifier as Downgrade. Play around with this intent classifier to see how it works.
You can also use text extractors to identify valuable pieces of data that already exist within NPS responses, most notably:
This model extracts the most relevant words or expressions within text. For example, Zoom could use a keyword extraction model to process NPS open-ended responses and identify the most common topics, e.g. videocall, lags, quality, etc. Type your own text into MonkeyLearn’s pre-trained model to see how it works.
Obtains names of people, companies, brands, and more. This can be useful for companies with a global presence, for example, by extracting different branch locations mentioned in your customer feedback, you can identify which one is being mentioned most often, and whether its being mentioned positively or negatively.
Use our pre-trained company entity extractor to quickly extract company entities from text.
NPS has become a game-changer for many companies, helping them gain valuable insights and make data-driven decisions.
While the NPS score is important, the most insightful feedback lies within customer responses to the follow-up question.
Customers’ responses tell businesses what needs to be improved and what’s working. As we mentioned above, there are some simple best practices you’ll need to follow, so that you can make the most out of your NPS responses.
We recommend starting with AI, since there are many online tools that make it straightforward to automatically analyze your unstructured text, plus you’ll gain more accurate insights from your NPS in seconds.
Sign up to MonkeyLearn for free and we’ll be happy to help you analyze your NPS responses with AI.
March 4th, 2020